The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the user to install the 'tpc' command-line interface using a piped shell command (curl -fsSL https://cli.promptingco.com/install.sh | bash). Executing remote scripts directly in the shell is an insecure practice that allows for arbitrary code execution without verification.
[EXTERNAL_DOWNLOADS]: The skill performs network requests to https://cli.promptingco.com to download installation scripts and retrieve experiment data.
[COMMAND_EXECUTION]: The workflows rely on multiple shell commands using the tpc CLI (e.g., tpc sim experiment results, tpc sim run logs) to gather information for analysis.
[DATA_EXFILTRATION]: Data from local experiment iterations, including agent transcripts and scores, is retrieved from or sent to the vendor's remote platform via the CLI tool.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by processing untrusted agent logs and transcripts from simulated experiments. Malicious content within these logs could potentially influence the agent's behavior during report generation.
Ingestion points: Data is ingested via tpc sim run logs and tpc sim experiment results as described in SKILL.md and workflows/analyze-experiment.md.
Boundary markers: No specific boundary markers or 'ignore' instructions are provided to isolate processed data from the agent's instructions.
Capability inventory: The agent can execute shell commands through the tpc CLI.
Sanitization: There is no explicit sanitization or filtering of the content retrieved from experiment logs before analysis.

analyze-experiment